What's Changed

This past month, we've been focused on day-2 operations and improving the usage of our recently launched features.

Event filtering

It is now possible to register filters for events that trigger tasks and workflows. For example, if you'd only like to trigger a workflow when the event email:received is sent from the user a@example.com to the user b@example.com, you could set up a filter for that:

:::py
hatchet.filters.create(
    workflow_id=event_workflow.id,
    expression='input.sender_email == "a@example.com"',
    scope='b@example.com',
    payload={
        'filterkey1': 'abcd'
    },
)

You can read more about filters here.

New Cloud Features

Managed Compute

• Our managed compute instances now support an additional Frankfurt region
• We've improved the usage of setting environment variables, and all environment variables are redacted in the UI by default

New Self-Hosted Features

Prometheus metrics

If you're a self-hosted user of Hatchet, you can now configure global prometheus metrics. We support the following metrics:

Read replicas

Self-hosted instances now support read replicas. Instead of placing additional load on your primary database with metrics and list queries, you can delegate these queries to a read replica. You can read about how to set up read replicas here.

Trace sampling

If you're running a high-volume Hatchet setup, it might not be desirable to store every single execution in the database. We've introduced trace sampling to reduce pressure in high-throughput scenarios.